• TDD Spotlight

    Performance of the HOMME Dynamical Core in the Simulation of Rainfall: Real-Planet Study with CAM3.5

    Researchers at NCAR and DOE laboratories have been working to ensure the CCSM is ready to fully utilize upcoming petascale computing platforms. The main bottleneck to petascale performance in the CCSM is the scalability of the atmospheric dynamical core. Team members at NCAR, SNL and ORNL have thus been focusing on the integration the Computational Science Section's High Order Method Modeling Environment (HOMME) into CAM, the atmospheric component of the CCSM. Initial work has been focused on HOMME's highly scalable and efficient spectral element dynamical core for solving the fluid dynamics equations necessary to study the atmosphere. The spectral element method in HOMME locally conserves both mass, tracer mass and energy on very general quadrilateral grids such as the quasi-isotropic cubed-sphere grid for the here. The dynamics are 4th order accurate and the non-oscillatory tracer advection is 3rd order accurate.

    Initial CCSM simulations have been completed, coupling CAM-HOMME (using the CAM 3.5.1 physics configuration) with the Community Land Model and prescribed ocean/ice conditions. Since rainfall is one of the most important components of the Earth's climate system, its simulation was examined in details. The simulated rainfall from HOMME has been compared with the observation and also with the Eulerian dynamical core (CAM-EUL). The area-averaged rainfall has been analyzed using monthly rainfall data for the tropics and extra-tropics. Simulation of the vertical thermodynamic structure of the troposphere is investigated and its connection to the simulated rainfall has been shown. The spatial distribution of rainfall, evaporation, large-scale moisture convergence, and wind are investigated in details. It is concluded that CAM-HOMME simulates various aspects of rainfall realistically. In certain respects e.g., partitioning of total rainfall between its components, CAM-HOMME performs better than CAM-EUL. To further improve its simulation capability, tuning effort with the CAM physics package is underway.